Determinants of the reliability of ultrasound tomography sound speed estimates as a surrogate for volumetric breast density

• Published in: Medical physics (Lancaster) Vol. 42; no. 10; pp. 5671 - 5678
• Main Authors: Khodr, Zeina G., Sak, Mark A., Pfeiffer, Ruth M., Duric, Nebojsa, Littrup, Peter, Bey‐Knight, Lisa, Ali, Haythem, Vallieres, Patricia, Sherman, Mark E., Gierach, Gretchen L.
• Format: Journal Article
• Language: English
• Published: United States American Association of Physicists in Medicine 01.10.2015
• Subjects: 60 APPLIED LIFE SCIENCES; acoustic tomography; Acoustical measurements; Adult; Aged; ANIMAL TISSUES; Biological material, e.g. blood, urine; Haemocytometers; BIOMEDICAL RADIOGRAPHY; Breast - cytology; breast cancer; breast density; cancer; change in breast density; Density measurement; Digital computing or data processing equipment or methods, specially adapted for specific applications; Digital mammography; Female; Film mammography; Humans; Image analysis; Image data processing or generation, in general; Image quality; MAMMARY GLANDS; mammography; medical image processing; Medical image quality; Middle Aged; NEOPLASMS; reliability; reproducibility; Reproducibility of Results; SOUND WAVES; Speed of sound; Tomography - methods; tumours; ultrasonic imaging; Ultrasonic Waves; Ultrasonographic imaging; ULTRASONOGRAPHY; Ultrasound Physics; ultrasound tomography
• ISSN: 0094-2405; 2473-4209; 2473-4209
• DOI: 10.1118/1.4929985

Purpose: High breast density, as measured by mammography, is associated with increased breast cancer risk, but standard methods of assessment have limitations including 2D representation of breast tissue, distortion due to breast compression, and use of ionizing radiation. Ultrasound tomography (UST) is a novel imaging method that averts these limitations and uses sound speed measures rather than x‐ray imaging to estimate breast density. The authors evaluated the reproducibility of measures of speed of sound and changes in this parameter using UST. Methods: One experienced and five newly trained raters measured sound speed in serial UST scans for 22 women (two scans per person) to assess inter‐rater reliability. Intrarater reliability was assessed for four raters. A random effects model was used to calculate the percent variation in sound speed and change in sound speed attributable to subject, scan, rater, and repeat reads. The authors estimated the intraclass correlation coefficients (ICCs) for these measures based on data from the authors’ experienced rater. Results: Median (range) time between baseline and follow‐up UST scans was five (1–13) months. Contributions of factors to sound speed variance were differences between subjects (86.0%), baseline versus follow‐up scans (7.5%), inter‐rater evaluations (1.1%), and intrarater reproducibility (∼0%). When evaluating change in sound speed between scans, 2.7% and ∼0% of variation were attributed to inter‐ and intrarater variation, respectively. For the experienced rater's repeat reads, agreement for sound speed was excellent (ICC = 93.4%) and for change in sound speed substantial (ICC = 70.4%), indicating very good reproducibility of these measures. Conclusions: UST provided highly reproducible sound speed measurements, which reflect breast density, suggesting that UST has utility in sensitively assessing change in density.